The helm charts already do this, it is a change for the kubectl/yaml
examples, which have been using NodePort. NodePort has been
unnecessary since we've automated testing by using port-forward to
the Service. This brings the two different deployment methods more
in line.

We will add tests for LoadBalancer and NodePort Services.

The docs are now even more out of sync with the actual example/test
code, since they still reference the NodePort type.

Previously the status was only updated when the sync to implement an
Ingress was successful. This will update the status due to changes
in addresses (IPs or hosts) assigned to the Service.

Also, the Service info sources for Ingress loadBalancer status depend
on the Service type, mostly following the nginx ingress implementation.

The addresses for this array are taken from the public names and/or
IPs in the spec for Service(s) that expose the Ingress. These are
identified as:

- in the same namespace as the admin Service
- have the label viking.uplex.de/svc=public
- have the same selectors as the admin Service
- type is one of ClusterIP, NodePort or LoadBalancer

The label viking.uplex.de/svc is only required if the Ingress status
update is required. For example to use a tool like ArgoCD, or if
the cloud provider requires it.

Set the label in the Service template for the viking-service chart.

No longer needed since we've set the Pods to ready regardless of
whether an Ingress has been configured.

The controller executable is the only artifact needed for its
container.

The templatedir CLI argument is removed.

This will be done incrementally for all templates, making the
controller executable wholly self-contained.

Closes #38

Service spec.ports maps to the public TLS port. First just for
kubectl/yaml deployments.

Addresses #38

Addresses #38

Addresses #38

Initially for kubectl/yaml deployments.

While here, ensure that /etc/varnish is world-readable, since it's
been reported that it might not be. Evidently not always, but setting
the permissions is never wrong.

Addresses #38

in vcl_backend_response {} and vcl_backend_error {} we test if
`bereq.backend == vk8s_cluster.backend(resolve=LAZY)`. With the
current code, this condition will never evaluate true, because
`resolve=NOW` will make the shard director return the "real" (VBE)
backend.

With this patch, the net effect, the chosen cluster varnish backend,
should still be the same, except for the rare race event that the
backend becomes unhealthy between the time the vcl code executes and
when the backend connection is made. We already account for this race
by setting `req.hash_ignore_busy` in vcl_revc {}.